Journal of Analytical Toxicology, Vol.

27, May/June 2003

Evaluation of Immunochemical Drug Screenings of Whole Blood Samples. A Retrospective Optimization of Cutoff Levels after Confirmation-Analysis on GCMS and HPLCDAD
Lars Kroener, Frank Musshoff, and Burkhard Madea
Institut of Legal Medicine, Bonn, Germany

Abstract
Four commonly used immunoassay kits were evaluated for their efficiency in screening for drugs of abuse in whole blood. Six groups of illicit drugs (opiates, cannabinoids, amphetamines, cocain and benzoylecgonine, benzodiazepines, and methadone) were determined by using the homogenous assays ADx and CEDIA DAU and compared with the results produced by means of the inhomogenous assays MTP and Pyxis 24. The measured 86 blood samples were taken from authentic routine analyses between February and September, 2000. Chromatographic confirmation analyses were carried out in all cases (positive and negative immunochemical pretesting). The cutoff levels were retrospectively optimized to reduce false-negative results with priority. Furthermore, false-positive pretests were minimized in order to decrease laboratory work under economical aspects. Specificity and sensitivity were determined for each parameter and assay. For the ADx assay, specificities of 54% (cannabinoids) to 97% (cocaine and metabolite) and sensitivities of about 67% (amphetamine class) to 94% (opiates) were found. The CEDIA assay revealed specificities of 77% (methadone) up to 100% (benzodiazepines) and 7596% sensitivities for amphetamines and opiates. The MTP immunoassay resulted in specificities of 52% (methadone) to 95% (opiates, cocain, and metabolite) and sensitivities of 92% (amphetamines) up to 100% (methadone). The evaluation of the Pyxis 24 resulted in specificities of 7096% (benzodiazepines and amphetamines) and sensitivities of 75% (amphetamines) up to 100% (cannabinoids and methadone), respectively. In conclusion, the microtiterplate immunoassays revealed higher sensitivities and have proved to be at an advantage detecting the lowest concentrations of drugs. However, especially for clinical applications in emergency cases with acute intoxications, when screening results are urgently required, homogenic assays such as ADx or Cedia provide preferable alternatives with faster and easier handling.

Introduction
Drug abuse is an increasing problem in modern society. It

leads to a rapidly growing number of clinical and forensic analyses. Therefore, it is necessary to apply efficient and rapid analytical methods for screening purposes that provide the highest sensitivities and specificities. The commonly used chromatographic methods in toxicological analysis gas chromatography (GC), gas chromatographymass spectrometry (GCMS), highperformance liquid chromatography (HPLC), and thin layer chromatography (TLC), are reliable, specific, and sensitive, but for rapid screenings of biological specimens in emergency cases, they are too complicated in their performance and much too time consuming. The development of immunochemical tests for the determination of drugs of abuse was an important progress in analytical toxicology. Initially developed for analyses of urine samples, it was soon of important practical interest to adapt immunoassay testing to serum or whole blood samples. But compared with the analysis of urine, homogenous immunoassays of turbid body fluids require sample preparation and cleanup steps to precipitate protein and hemoglobine contents that would interfere with the detection of the analytes. Several workgroups published useful procedures with organic solvents as precipitation reagents to generate an almost clear supernatant for analytic measurements. Liquidliquid extraction with chloroformethanol was used (1) in combination with the EMIT-dau method, as well as protein precipitation with methanol (2,3,4) or acetone (5,6) prior to EMIT-dau or FPIA immunoassays. The CEDIA DAU urine immunoassay was also used successfully for the determination of cannabinoids in whole blood after protein precipitation with acetone (7) and for screening purposes in whole blood and serum without any pretreatment of the sample (8). Microtiterplate immunoassays require less sample preparation and volume. The enzyme-linked immunosorbant assays use antibodies immobilized on the surface of a microtiterplate. During the incubation time after the addition of the sample, there is a competition between the drugs and drugenzyme conjugates added to the sample in binding to the sorbent. All

Reproduction (photocopying) of editorial content of this journal is prohibited without publishers permission.

205

Journal of Analytical Toxicology, Vol. 27, May/June 2003

nonbonded drugs and blood contents were washed away in a subsequently occuring rinsing step and may not cause any interference. Commercially available microtiterplate assays were evaluated for different drug classes (9,10) and compared with commonly used homogenous immunoassays (1114). Inhomogenous assays proved to be more sensitive with a higher specificity. In the recent study, the pretest results of two homogenous assays, ADx and CEDIA DAU, and two microtiterplate assays, Mahsan MTP and PYXIS 24, were compared with the results attained by means of GCMS and HPLCdiode array detection (DAD) for 86 whole blood specimens. Retrospectively, the cutoff levels of the immunoassays were optimized for each of the tested compounds.

used for the study. All calibration standards were also supplied by the manufacturer. All instrument settings were the same as the manufacturers recommended values. The immunoassay kits were purchased from Abbott GmbH Diagnostika.
CEDIA DAU

The CEDIA assays were performed on a Hitachi 911 automatic analyzer (Boehringer Mannheim/Hitachi, Mannheim, Germany). Parameters used for the analyzer settings were the same as those recommended by the manufacturer. The immunoassay kits were obtained from Microgenics GmbH (Passau, Germany).
MAHSAN MTP

Materials and Instrumentation

Materials

All drug standards and deuterated internal standards (IS) were purchased from Promochem (Wesel, Germany). All solvents and buffer solutions were puchased from Merck (Darmstadt, Germany) in HPLC grade. N -Methyl- N -trimethylsilyltrifluoroacetamide (MSTFA) and N-methyl-bis-(trifluoroacetamide) were puchased from Pierce (Perbio Science, Bonn, Germany) and Macherey-Nagel (Dueren, Germany). Solid-phase extractions (SPE) were carried out on Chromabond-DrugColumns (Macherey-Nagel).

The extinctions of the MTP were measured on a spectral microtiterplate photometer (Tecan AG, Crailsheim, Germany). All analyses were performed using parameters recommended by the manufacturer. The microtiterplates and reagents were purchased from Mahsan-Diagnostika (Reinbek, Germany).
Pyxis 24

A fully automated analyzer CODA (Biorad GmbH, Munich, Germany) was used for all assays. All settings were the same as recommended by the manufacturer. The microtiterplates and reagents were purchased from Bio-Rad GmbH.
GCMS

An Agilent 5890 Series II Plus GC (Waldbronn, Germany) coupled to a 5972 mass selective detector was used for analysis. Immunoassay instrumentation Data acquisition and data analysis were performed on the For detailed instrument settings and analytical characterisChemstation software. For all analyses, fused-silica capillary tics, see Table I. columns HP-5MS (30 m 0.25 mm, 0.25-m film thickness) ADx (Agilent) were used with a flowrate of 1 mL helium/min after An ADx analyzer (Abbott GmbH, Wiesbaden, Germany) was splitless injection at an injector temperature of 280C. The amphetamine group was chromatographed with a starting temperature of 80C, held for 1 Table I. Instrument Settings and Analytical Characteristics min, then increased 10C/min up to 280C, Immunoassays and a held for 2 min. For the separation of coAnalytes caine, benzoylecgonine, and opiates, the CAN COC OP AMP BENZ METH column temperature was set at 130C for 1 min, raised 10C/min up to 290C, and held ADx 0100 (20) 01500 (20) 0400 (15) 0500 (25) 0100 (0) 01000 (20) for 1 min. For the cannabinoids, a temperature CEDIA DAU 050 (20) 01000 (20) 0600 (10) 01200 (25) 0500 (50) 01000 (50) program with an initial temperature of 80C MAHSAN MTP 050 (2) 0300 (10) 0500 (10) 0500 (25) 0500 (10) 0100 (5) was used with a heating rate of 13C/min up to PYXIS 24 050 (3) 0300 (30) 0500 (10) 0500 (20) 0500 (20) 0100 (20) 250C, held for 3 min, and then an attached Calibration range and cutoff levels (in parentheses) (ng/mL) heat-out program with a heating rate of 20C/min up to 300C was held for 5 min.
Confirmation methods Analytes GCMS CAN LOD LOQ 0.3 1.3 COC 0.25 1 OP 0.5 2.1 AMP 1.7 7.1 HPLCDAD BENZ 1.5 6.1 METH 1.4 5.8

HPLC

* Abbreviations: CAN = cannabinoids, COC = cocaine and cocainemetabolite, OP = opiates, AMP = amphetamines, BENZ = benzodiazepines, and METH = methadone.

A Shimadzu setup equipped with an LC-6A pump system and an SPD-M10AVP diode array detector (200340 nm wave length) was used for HPLC analyses at a flow rate of 0.5 mL/min. A LiChroCART column with LiChrosorb-RP8 phase (Merck) was used with phosphate buffer 0.1M pH 2.3acetonitrile (2:1) as the mobile phase.

206

Journal of Analytical Toxicology, Vol. 27, May/June 2003

Methods
Analytical characteristics are given in Table I. All methods were validated according to IS (ISO 17025). The methods presented below are the standard methods used in our laboratory.
Amphetamine class

the IS, 1 mL of blood or serum was extracted with 5 mL of chlorobutane. Evaporated to dryness and reconstituted with 100 L of the mobile phase was 4 mL of the organic layer, and a 50-L aliquot was analyzed by means of HPLC with DAD.
Methadone

After the addition of 100 L of the deuterated IS mixture (200 ng/mL of amphetamine-d5, methamphetamine-d11, methylenedioxyamphetamine-d5, methylendioxymethamphetamine-d5, and methylenedioxyethylamphetamine-d5) and 30 L of 2N sodium hydroxide solution to 200 L of blood or serum the mixture was extracted with 500 L of n-hexane. Subsequently, 160 L of the organic layer were transferred into a 2-mL vial. After the addition of 40 L of MBTFA [Nmethyl-(N,N-bistrifluoroacetamide)] and trifluoroacetylation at 70C for 20 min, the solution was analyzed by GCMS in selected ion monitoring (SIM) mode. The ions monitored were 118, 140, and 91 amu for amphetamine; 154, 110, and 118 amu for methamphetamine; 135, 162, and 275 amu for methylenedioxyamphetamine; 154, 135, and 110 amu for methylendioxymethamphetamine; and 168, 140, and 303 amu for methylenedioxyethylamphetamine.
Opiates, cocaine, and benzoylecgonine

Blood or serum in the amount of 1 mL was adjusted with 1 mL of phosphate buffer to pH 7 and subsequently extracted with 5 mL of dichloromethane. Evaporated to dryness and reconstituted with 100 L of the mobile phase was 4 mL of the organic layer. A 50-L aliquot was analyzed by means of HPLC with DAD.
ADx immunoassay

To 1 mL of whole blood or serum, 1 mL acetone was added and vigorously mixed. After centrifugation (5 min at 11000 rpm), the supernatant was analyzed directly. The calibration and control serum samples were analyzed within every sequence.
CEDIA DAU immunoassay

After the addition of 100 L of the deuterated IS mixture (200 ng/mL of cocaine-d3, benzoylecgonine-d3, codeine-d3, dihydrocodeine-d3, morphine-d3, and 6-acetylmorphine-d3) and 800 L of acetone to 800 L of blood or serum, the solution was mixed intensively for protein precipitation. The supernatant was evaporated to dryness and reconstituted with 4 mL of phosphate buffer (pH 6). The resulting solution was cleaned up by mixedmode SPE, evaporated to dryness, and silylated with a mixture of 70 L MSTFA, 30 L pyridine, and 100 L of iso-octane for 20 min at 90C. An aliquot was injected into the GCMS operating in SIM mode. The ions monitored were 182, 303, and 272 amu for cocaine; 240, 361, and 346 amu for benzoylecgonine; 373, 236, and 282 amu for dihydrocodeine; 371, 234, and 343 amu for codeine; 429, 236, and 401 amu for morphine; and 399, 340, and 287 amu for 6-acetylmorphine.
Cannabinoids

Mixed intensively were 500 L of whole blood or serum and 1 mL acetone. After sharp centrifugation (5 min at 11000 rpm) and the addition of 1 drop of 0.1M HCl, the supernatant was evaporated to dryness and reconstituted in 1 mL of phosphate buffer solution (pH 7). The resulting solution was analyzed together with serum control samples and calibrators automatically in a Hitachi 911 analyzer.
MTP immunoassay

Whole blood or serum in the amount of 25 L was pipetted undiluted together with 100 L of the enzyme conjugate (horseradish-peroxidase/coupled to the drug of interest) into the antibody-coated microtiterplate wells. After an incubation period of 30 min, the microtiterplate was washed with water six times and beaten out at the workbench to remove the residues. Added to the wells and incubated for 30 min in darkness was 100 L of a substrate solution (3,3',5,5'-tetramethylbenzidin). The reaction was stopped by the addition of sulfuric acid, and the microtiterplate was analyzed by the TECAN reader with easy WIN fitting software. The calibrators provided by the manufacturer and control serum samples were analyzed within every batch.
PYXIS 24 immunoassay

After the addition of 100 L of a deuterated IS mixture (100 ng/mL 9-tetrahydrocannabinol-d3, 100 ng/mL 11-hydroxy-9tetrahydrocannabinol-d3, and 200 ng/mL 11-nor-9-carboxy-9tetrahydrocannabinol-d9), 1 mL of blood was extracted with 5 mL of n-hexaneethyl acetate (9:1). After the extraction, the blood was adjusted to pH 4 by adding 0.1M hydrochloric acid and extracted one more time with 5 mL of n-hexaneethylacetate. The combined organic layers were evaporated to dryness and silylated with a mixture of 70 L MSTFA, 30 L of pyridine, and 100 L of iso-octane for 20 min at 90C. An aliquot was analyzed by GCMS in SIM mode. The ions monitored were 371, 386, and 303 amu for 9-tetrahydrocannabinol; 371, 474, and 459 amu for 11-hydroxy-9-tetrahydrocannabinol; and 371, 473, and 488 amu for 11-nor-9-carboxy-9-tetrahydrocannabinol.
Benzodiazepines

A 500-L aliquot of whole blood or serum was transferred into a 2-mL tube and placed into the analyzer. No predilution of the sample was required. All analysis steps were executed automatically according to the manufacturers recommendations and were, overall, comparable with the MAHSAN procedure described previously.
Optimization of the cutoff levels

After addition of 50 L of a brotizolam solution (1 g/mL) as

The retrospective optimization of the cutoff levels was performed on a personal computer using Microsoft Excel 97. By varying the threshold values of the immunoassays and comparing the effects automatically with the results of the confirmation analyses, optimum cutoff levels were calculated. The primary object was the diminishing of false-negative results and an enhanced sensitivity by lowering the cutoff. The number of false-negative results should be as low as possible. Secondarily,

207

Journal of Analytical Toxicology, Vol. 27, May/June 2003

the false-positive results were minimized by adjusting the cutoff to a higher level to obtain a higher specificity without increasing the number of false-negative results. If the number of false-negatives began to raise, the cutoff was set back to the next lower value. Sensitivity and specificity were calculated according to the following formulas: Sensitivity = TP/(TP + FN) Specificity = TN/(TN + FP) Eq. 1 Eq. 2

Because sensitivity and specificity are probabilities, the standard error must be calculated according to the following formula: SEp = pq/n Eq. 3

where p is the sensitivity or specificity, q = 1 p, and n is the sample size (15).

Results
Confirmation analysis

below 1 ng/mL were considered negative because of the mismatching qualifier ion ratios and negative immunoassay results. For cocaine and benzoylecgonine, 27 samples tested positive. Benzoylecgonine, without any evidence for cocaine, appeared to be positive in 14 cases. The concentrations for cocaine were determined in a range between 1 and 300 ng/mL of whole blood and benzoylecgonine between 10 and 1800 ng/mL. Thirtyone specimens were positive for opiates. Free morphine and codeine were detected mostly in amounts between 1 and 10 ng/mL. The confirmed compounds of the amphetamine class were amphetamine, methylenedioxyamphetamine, and methylenedioxymethylamphetamine, most often with concentrations higher than 100 ng/mL. MDA as a metabolite of MDMA appeared to be positive predominantly in concentrations between 10 and 25 ng/mL. Most of the methadone-positive samples showed concentrations higher than 50 ng/mL. Diazepam and desmethyldiazepam were confirmed commonly with concentrations between 200 and 2000 ng/mL. The highest value found was for diazepam with a concentration of 4600 ng/mL. Oxazepam and temazepam were determined in a range from 10 to 100 ng/mL. All other detected benzodiazepine derivatives were found less often, mostly at concentration levels between 50 and 250 ng/mL. Only 5 of 86 specimens were negative for all measured analytes.
Cannabinoid immunoassays

The results of the confirmation analyses of 86 blood samples are summarized in Table II. Most cannabinoid-positive samples were determined with tetrahydrocannabinol and tetrahydrocannabinol carboxylic acid concentrations between 1 and 10 ng/mL. Seven blood samples tested positive only for THC-COOH within a range of 2 to 50 ng/mL. Five cases with concentrations

The ADx immunoassay yielded 56 positive results at a cutoff level of 20 ng/mL. Overall, 4 false-negative and 22 false-positive results were recognized after confirmation analysis by means of GCMS (Table III). Reducing the cutoff value did not contribute to a better result. All false-negative analyses produced by the assay were not affected by anTable II. Results of 86 Blood Samples Analyzed by GCMS and HPLCDAD alyte concentrations near the limits of detection (LOD), but more likely by matrix effects. 110 1025 2550 50100 > 100 Lower cutoff levels led to more false-positive Pos ng/mL ng/mL ng/mL ng/mL ng/mL results and a decreased specificity. The CEDIA assay produced 8 false-negative THC 26 24 2 0 0 0 and 4 false-positive results using a cutoff level of THC-OH 18 17 1 0 0 0 THC-COOH 38 13 8 9 5 3 20 ng/mL. Diminishing the cutoff to 5 ng/mL Total cannabinoids 38 (Table IV) led to an increased sensitivity with only 2 false negative results. Twenty-four falseCOC 13 5 3 1 2 2 positives, as much as true-positives, decreased BE 27 0 0 4 2 21 the specificity of the assay to 50%. A correlation Total COC and BE 27 between the concentration of THC, THC-OH, or THC-COOH and false immunoassay results MORPH 28 12 8 8 0 0 could not be recognized so that matrix effects COD 23 17 5 1 0 0 were supposed to be the cause. DHC 2 0 0 1 0 1 The MAHSAN MTP assay only produced 1 6-MAM 1 1 0 0 0 0 false-negative analysis corresponding to a Total opiates 31 sensitivity of 97% at a cutoff level of 2 ng/mL. AMPH 7 1 1 1 0 4 Increasing the cutoff to 20 ng/mL MAMPH 0 0 5 1 0 0 improved the specificity from 27% to 63% or MDA 6 0 0 0 0 0 35 to 18 false-positive assays, respectively. MDMA 7 0 1 0 0 6 The PYXIS 24 showed no false-negative reMDEA 0 0 0 0 0 0 sults and 100% sensitivity at a cutoff value of 3 Total amphetamines 12 ng/mL. The 60% specificity was improved to Methadone 25 0 3 4 8 10 73% by setting the cutoff up to 10 ng/mL. Benzodiazepines 40 In some of the cases with false-positive mi-

208

Journal of Analytical Toxicology, Vol. 27, May/June 2003

by retrospective optimization of the cutoff levels. Thirteen false-positive and one false-negative results were found at a cutoff level of 20 ng/mL. In whole blood specimens, in which only benzoylecgonine was found, the quantitative results of the assay corresponded in most of the cases with the confirmation by means of GCMS. All false-positive results occurred in a range of 20 to 30 ng/mL. A false-negative analysis was confirmed with a concentration of 50 ng/mL benzoylecgonine. Cocaine and benzoylecgonine immunoassays The MAHSAN MTP showed a high sensitivity of 93% applying The ADx assay with the recommended cutoff level of 20 a cutoff level of 10 ng/mL. Less false-positives could be obng/mL operated near the optimum with 3 false-negative retained by increasing the threshold up to 30 ng/mL with a sults. The specificity was enhanced by increasing the cutoff to number of 4 false-positive results and a corresponding speci35 ng/mL. A decreased number of false-positives (three instead ficity of 93%. Two false-negative assays were detected with conof five) could be attained. centrations of 31 and 1070 ng/mL of benzoylecgonine after The results of the CEDIA DAU assay could not be improved GCMS analysis. The PYXIS 24, which operated at a cutoff value set to 30 ng/mL, produced 89% sensiTable III. Agreement Between Immunochemical Results and Confirmation tivity and 97% specificity. At a lower cutoff of 8 Analysis Without Optimization of Cutoff Levels ng/mL no false-negative results were detected, but 14 false-positive assays occurred, on the A. Number of positive and negative ADx immunoassay results compared with confirmation analysis other hand. The false-positive results were deADx immunoassay termined by the assay in a range of 10 to 50 Analyte (cutoff level) Confirmation ng/mL.
analysis CAN (20) + COC, BE (20) + 24 4 3 55 OP (15) + AMP (25) + BENZ (0) + METH (20) + 23 5 2 56

crotiterplate assays, the subsequently carried out GCMS analyses showed possible traces of THC-COOH, but because of the mismatching qualifier ion ratios or the lack of the qualifiers, the presence of cannabinoids was not sufficiently evidential. The only false-negative MTP assay occured in a case with less than 2 ng/mL of THC and 5 ng/mL of THC-COOH, thus four times higher than the LOD given by the manufacturer.

Opiate immunoassays

34 22

4 26

29 8

2 47

8 10

4 64

37 16

3 30

B. Number of positive and negative CEDIA immunoassay results compared with confirmation analysis CEDIA DAU immunoassay Analyte (cutoff level) Confirmation analysis CAN (20) COC, BE (20) OP (10) AMP (25) BENZ (50) METH (50) + + + + + + + 30 4 0 44 24 13 3 46 30 8 1 47 10 7 2 67 38 0 2 46 23 14 2 47

C. Number of positive and negative PYXIS 24 immunoassay results compared with confirmation analysis PYXIS 24 immunoassay Analyte (cutoff level) Confirmation analysis CAN (3) COC, BE (30) OP (10) AMP (20) BENZ (20) METH (20) + + + + + + + 38 19 0 29 24 2 3 57 30 8 1 47 10 8 2 66 38 11 2 35 23 4 2 57

D. Number of positive and negative MAHSAN MTP immunoassay results compared with confirmation analysis MAHSAN MTP immunoassay Analyte (cutoff level) Confirmation analysis CAN (2) COC, BE (10) OP (10) AMP (25) BENZ (10) METH (5) + + + + + + + 37 35 1 13 25 28 2 31 29 16 2 39 9 4 3 70 36 7 4 39 25 46 0 15

* Abbreviations: CAN = cannabinoids, CPC = cocaine and cocainemetabolite, OP = opiates, AMP = amphetamines, BENZ = benzodiazepines, and METH = methadone.

The opiate assay of the ADx produced eight false-positive and two false-negative analyses with cutoff settings recommended by the manufacturer. The false-negative assays were found at analyte concentrations below 2 ng/mL for morphine and 3 ng/mL of codeine, respectively. Applying a cutoff level of 20 ng/mL instead of 15 ng/mL, decreased the false-positive results from 8 to 5. A sensitivity of 94% and a specificity of 85% were achieved by means of the CEDIA assay operating at a cutoff level of 10 ng/mL. Enhancements could only be obtained by applying a cutoff level of 5 ng/mL, which decreased the number of false-negative assays to 1 with 17 false-positives. The false-positive assays were mainly given out with less than 15 ng/mL. The false-negative result was confirmed with 11 ng/mL. After increasing the cutoff level of the MAHSAN MTP from 10 ng/mL up to 35 ng/mL, the specificity was enhanced from 71% to 85%. The sensitivity remained at 94% and could not be enhanced further. In the other 2 cases of false-negative assays, analyte concentrations of 24 ng/mL morphine and 48 ng/mL morphine combined with 12 ng/mL of codeine were determined by GCMS. The performance of the PYXIS assay employing the manufacturers settings was already at an optimum. Only 1 false-negative with 3 ng/mL of morphine and 8 false-positive results were found.

209

Journal of Analytical Toxicology, Vol. 27, May/June 2003

Amphetamine immunoassays

With a cutoff level of 25 ng/mL as recommended by the manufacturer, 67% sensitivity and 86% specificity were found with the ADx assay. A slightly improved specificity could be obtained with a cutoff set to 30 ng/mL. The number of false-positives decreased from ten to eight. Between the false-negative results, no relationship could be recognized. Amphetamine and MDMA were determined in these cases with concentrations of more than 100 ng/mL. The CEDIA assay produced 4 false-negatives and 7 false-positives before setting the cutoff level from 25 to 20 ng/mL. After optimization, 3 false-negative assays with concentrations of more than 100 ng/mL of amphetamine and 25 ng/mL of MDA were detected. Eight false-positive results were found. The relatively low sensitivity of 75% found with the MAHSAN

MTP at a cutoff level of 25 ng/mL was increased to 92% by setting the cutoff value up to 20 ng/mL. Specificity remained at remarkable 95%. One false-negative and four false positives were found after optimization. With a cutoff level set to 20 ng/mL, similar to the manufacturers recommendations, 83% sensitivity and 89% specificity were found caused by 2 false-negatives (16 ng/mL of amphetamine and 10 ng/mL of MDMA) and 8 false-positive results. A cutoff level of 25 ng/mL improved the specificity to 91% with 7 false-positives.
Benzodiazepine immunoassays

The benzodiazepine assay of ADx yielded a sensitivity of 93%, which could not be improved because the cutoff was set already to 0 ng/mL. The specificity was 65% because of 16 false-positive results. A cutoff level of 5 ng/mL improved the specificity to 78% with 10 positive, not conTable IV. Agreement Between the Immunochemical Results and Confirmation Analysis After Optimization of the Cutoff Levels firmable results. False-positive assays occurred only in cases when diazepam in concentrations A. Number of positive and negative ADx immunoassay results compared with confirmation analysis higher than 1000 ng/mL and nordiazepam in ADx immunoassay traces were present. Analyte (cutoff level) The CEDIA assay had an optimum specificity Confirmation analysis CAN (20) COC, BE (35) OP (20) AMP (30) BENZ (5) METH (20) of 100% at the recommended setting level for + + + + + + the cutoff value. Only the sensitivity could be increased by applying a higher cutoff level of 50 + 34 4 24 3 29 2 8 4 37 3 23 2 instead of 20 ng/mL. 22 26 3 56 5 50 8 66 10 36 5 56 The MAHSAN assay was improved by setting the cutoff value from 10 to 5 ng/mL. The sensiB. Number of positive and negative CEDIA immunoassay results compared with confirmation analysis tivity rose from 90% to 93%, but the specificity CEDIA DAU immunoassay was decreased from 85% to 61%. Eighteen Analyte (cutoff level) Confirmation false-negative results were found for diazepam analysis CAN (5) COC, BE (21) OP (40) AMP (20) BENZ (20) METH (50) and nordiazepam concentrations of 27 and 12 + + + + + + ng/mL and for temaze-pam concentrations of 92 ng/mL with traces of oxazepam. + 36 2 26 1 29 2 9 3 37 3 24 1 With 95% sensitivity and 76% specificity, 24 24 13 46 3 52 8 66 0 46 14 47 the PYXIS 24 assay turned out to be a little C. Number of positive and negative PYXIS 24 immunoassay results compared with confirmation more reliable. Two false-negatives occurred analysis in one case with 21 ng/mL of diazepam and PYXIS 24 immunoassay another with diazepam, nordiazepam, 7Analyte (cutoff level) aminoflunitraze-pam, and flunitrazepam in Confirmation concentrations between 10 and 50 ng/mL. analysis CAN (10) COC, BE (22) OP (10) AMP (25) BENZ (10) METH (15)
+ + 38 13 0 35 + 27 18 0 41 + 30 8 1 47 + 10 7 2 67 + 38 14 2 32 + 25 5 0 56

Methadone immunoassays

D. Number of positive and negative MAHSAN MTP immunoassay results compared with confirmation analysis MAHSAN MTP immunoassay Analyte (cutoff level) Confirmation analysis CAN (20) COC, BE (30) OP (35) AMP (20) BENZ (5) METH (35) + + + + + + + 37 18 1 30 25 4 2 55 29 8 2 47 11 4 1 70 37 18 3 28 25 29 0 32

* Abbreviations: CAN = cannabinoids, COC = cocaine and cocainemetabolite, OP = opiates, AMP = amphetamines, BENZ = benzodiazepines, and METH = methadone.

A sensitivity and specificity of 92% were obtained applying a recommended cutoff level of 20 ng/mL with the ADx assay. Two falsenegatives appeared with methadone concentrations below 20 ng/mL. A further optimization was impossible. A sensitivity of 96% and 77% specificity were found for the CEDIA assay at a cutoff value of 50 ng/mL. Only 1 false-negative was found with a determined methadone concentration of 19 ng/mL. No false-negative results were found employing the MAHSAN MTP at a cutoff level of 5 ng/mL, but a specificity of only 25% was ob-

210

Journal of Analytical Toxicology, Vol. 27, May/June 2003

tained. A higher cutoff value of 35 ng/mL increased the specificity to 52%, respectively. A sensitivity of 100% was found for the PYXIS 24 assay, too. However, a higher specificity of 93% before and 92% after optimization of the cutoff value were found.

Conclusion
The results of this study expressed in sensitivity and specificity values before and after optimization of the cutoff levels are

shown in Table V. The optimized cutoff values are used for clinical and forensic casework in our laboratory with great success. False-positive screenings were minimized, and no false-negative results were detected in randomly carried out GCMS and HPLC analyses. It is shown that most of the sensitivities are greater than 90% after optimization of the cutoff levels. The fully automated PYXIS 24 showed the best results for sensitivity, reaching 100% for cannabinoids, methadone, cocaine, and benzoylecgonine. The MAHSAN MTP results were comparable to these. The number of false-negatives and false-positives were slightly

Table V. Sensitivities and Specificities of the Immunoassays Before and After Optimization of the Applied Cutoff Levels
A. Abbott ADx CAN Before optimization Cutoff (ng/mL) Sensitivity (%) Specificity (%) After optimization Cutoff (ng/mL) Sensitivity (%) Specificity (%) B. Microgenics CEDIA DAU CAN Before optimization Cutoff (ng/mL) Sensitivity (%) Specificity (%) After optimization Cutoff (ng/mL) Sensitivity (%) Specificity (%) C. MAHSAN MTP CAN Before optimization Cutoff (ng/mL) Sensitivity (%) Specificity (%) After optimization Cutoff (ng/mL) Sensitivity (%) Specificity (%) D. Biorad PYXIS 24 CAN Before optimization Cutoff (ng/mL) Sensitivity (%) Specificity (%) After optimization Cutoff (ng/mL) Sensitivity (%) Specificity (%) COC, BE OP AMP BENZ METH COC, BE OP AMP BENZ METH COC, BE OP AMP BENZ METH COC, BE OP AMP BENZ METH

20 89.5 5.0 54.2 8.1 20 89.5 5.0 54.2 8.1

20 88.9 6.0 93.2 4.8 35 88.9 6.0 94.9 4.2

15 93.5 4.4 85.5 6.3 20 93.5 4.4 90.9 5.2

25 66.7 13.6 86.5 9.9 30 66.7 13.6 89.2 9.0

0 92.5 5.3 65.2 9.5 5 92.5 5.3 78.3 98.2

20 92.0 4.3 91.8 4.3 20 92.0 4.3 91.8 4.3

20 78.9 6.6 91.7 4.5 5 94.7 3.6 50.0 8.1

20 96.3 3.6 78.0 8.0 21 96.3 3.6 78.0 8.0

10 93.5 4.4 85.5 6.3 40 93.5 4.4 94.5 4.1

25 66.7 13.6 90.5 8.4 20 75.0 12.5 89.2 9.0

50 90.0 6.0 100 0.0 20 92.5 5.3 100 0.0

50 96.0 3.1 77.0 6.6 50 96.0 3.1 77.0 6.6

2 97.4 2.6 27.1 7.2 20 97.4 2.6 62.5 7.9

10 92.6 5.0 52.5 9.6 30 92.6 5.0 93.2 4.8

10 93.5 4.4 70.9 8.2 35 93.5 4.4 85.5 6.3

25 75.0 12.5 94.6 6.5 20 91.7 8.0 94.6 6.5

10 90.0 6.0 84.8 7.2 5 92.5 5.3 60.9 9.8

5 100 0.0 24.6 6.8 35 100 0.0 52.5 7.9

3 100 0.0 60.4 7.9 10 100 0.0 72.9 7.2

30 88.9 6.0 96.6 3.5 22 100 0.0 69.5 8.5

10 96.8 3.2 85.5 6.3 10 96.8 3.2 85.5 6.3

20 83.3 10.8 89.2 9.0 25 83.3 10.8 90.5 8.4

20 95.0 4.4 76.1 8.5 10 95.0 4.4 69.6 9.2

20 92.0 4.3 93.4 3.9 15 100 0.0 91.8 4.3

211

Journal of Analytical Toxicology, Vol. 27, May/June 2003

Table VI. Mean Values for Sensitivities After Optimization of the Cutoff Levels
ADx Sensitivity (%) Specificity (%) 87 83 CEDIA 91 81 MAHSAN 95 75 PYXIS 96 80

2. 3. 4.

higher, even after optimization. Reasons for failure may be the analyte concentrations near the cutoff values as shown in some cases, but more likely matrix effects and errors caused by the operator, especially for the manually prepared MAHSAN MTP, may be the main causes. The optimized cutoff-values for the MAHSAN assay were comparable to those published by Kferstein et al. (13). Kauert et al. (14) determined the most practicable cutoff for cannabinoids at 2 ng/mL and for opiates at 30 ng/mL, respectively. For cannabinoids, 2 ng/mL in this study led to a poor specificity without any improvement of the sensitivity. Overall, it is nearly impossible to obtain reliable quantitative results by the evaluated immunoassays. Only in a few special cases, the quantitative immunoassay results were confirmed by further analyses. However, there was almost no correlation between the immunochemical pretests and confirmation analyses. Employing appropriate cutoff values, most of the assay results could have been improved. Less drug-positive specimens are missed by the pretests, and a decreased number false-positive pretests saves laboratory work and time. The homogenous ADx and CEDIA DAU assays are recommened for all purposes requiring reliable analytical results as fast as possible (i.e., in clinical emergency cases of intoxications, where often higher concentrations are found). Last, but not least, they are less expensive compared with microtiterplate assays. The heterogenous immunoassays provide overall high sensitivities, neccessary especially in forensic toxicology. Even in cases with a long period between incidence and blood sampling, lowest concentrations of drugs of abuse can be found, even if there is no observable effect on an accused person left. However, the difference between the CEDIA DAU and the evaluated microtiterplate immunoassays was small, as shown in Table VI, so that economical aspects may be decisive for toxicological laboratories in the procurement of immunochemical pretest equipments for routine drug screenings.

5. 6.

7.

8.

9. 10.

11.

12. 13.

14.

15.

an indirect homogenous enzyme immunoassay. J. Forensic Sci. 23: 292303 (1978). H.W. Peel and B.J. Perrigo. Detection of cannabinoids in blood using EMIT. J. Anal. Toxicol. 5: 16567 (1981). W. M. Asselin, J. M. Leslie, and B. McKinley. Direct detection of drugs of abuse in whole hemolyzed blood using the EMIT d.a.u. urine assay. J. Anal. Toxicol. 12: 20715 (1988). T. Daldrup and F. Musshoff. Forensische Analytik. Drogen und Arzneimittel. In Analytiker Taschenbuch, H. Gnzler, A. M. Bahadir, R. Borsdorf, K. Danzer, W. Fresenius, W. Huber, I. Lderwald, G. Schwedt, G. Tlg, and H. Wisser, Eds. Springer Verlag, Berlin/Heidelberg, Germany, 1995, 13, pp 183233. L.J. Lewellen and H.H. McCurdy. A novel procedure for the analysis of drugs in whole blood by homogenous enzyme immunoassay. J. Anal. Toxicol. 12: 20715 (1988). M. Bogusz, R. Aderjan, G. Schmitt, E. Nadler, and B. Neureither. The determination of drugs of abuse in whole blood by means of FPIA and EMIT-dau immunoassaya comparative study. Forensic Sci. Int. 48: 2737 (1990). J. C. Cagle, H. McCurdy, Y.M. Pan, K.J. Ayton, W.H. Wall, and E.T. Solomons. Evaluation of the CEDIA DAU assays and the AxSym system for the analysis of cannabinoids in whole blood. J. Anal. Toxicol. 21: 213217 (1997). S. Iwersen-Bergmann and A. Schmoldt. Direct semiquantitative screening of drugs of abuse in serum or whole blood by means of CEDIA DAU urine immunoassay. J. Anal. Toxicol. 23: 247256 (1999). B.J. Perrigo and B.P. Joynt. Use of ELISA for the detection of common drugs of abuse in forensic blood samples. Can. Soc. Forens. Sci. J. 28: 261269 (1995). K.A. Moore, C. Werner, R.M. Zannelli, B. Levine, and M.L. Smith. Screening postmortem blood and tissues for nine cases of drugs of abuse using automated microplate immunoassay. Forensic Sci. Int. 106: 93102 (1999). V.R. Spiehler, I.B. Collison, P.R. Sedgwick, S.L. Perez, S.D. Le, and D.A. Farnin. Validation of an automated microplate enzyme immnoassay for screening of postmortem blood for drugs of abuse. J. Anal. Toxicol. 22: 573579 (1998). H. Kferstein and G. Sticht. Vergleich der MTP-immunoassays mit EMIT beim screening von blut auf drogen. Arch. Kriminol. 202: 165172 (1998). H. Kferstein and G. Sticht. Results of immunochemical and chromatographic investigations of blood samples in suspected drug abusefirst report: opiates, cocaine, cannabinoids and amphetamines. Rechtsmedizin 8:173177 (1998). S.W. Toennes and G. Kauert. Immunochemical screening of serum samples: comparison of the cannabinoid and opiate assays Mahsan MTP and Abbott FPIA controlled by gas chromatography-mass spectrometry. Rechtsmedizin 10: 7175 (2000). M.H. Zweig, E.R. Ashwood, R.S. Galen, R.H. Plous, and M. Robinowitz. Assessment of the clinical accuracy of laboratory tests using receiver operating characteristics (ROC) plots; Approved guidelin GP-10A. National Committee for Clinical Laboratory Standards (NCCLS) 15: 127 (1995).

References
1. E.L. Slightom. The analysis of drugs in blood, bile and tissue with Manuscript received November 9, 2001; revision received May 29, 2002.

212